cellhashR

cellhashR performs demultiplexing of cell hashing data to classify droplets by sample origin and improve multiplexing accuracy in droplet-based single-cell sequencing.

Key Features:

• Bimodal Flexible Fitting (BFF) algorithms: Implements BFF demultiplexing algorithms, namely BFFcluster and BFFraw, that leverage the assumption of bimodal barcode count distributions.
• Integrated quality control (QC): Provides QC features to assess data integrity throughout the demultiplexing process.
• Robustness to data variability: Includes a tunable BFFcluster algorithm optimized for robust performance on poorly behaved input data.

Scientific Applications:

• Single-cell transcriptomics multiplexing: Enables cell hashing-based multiplexing to increase capacity on droplet-based platforms and reduce sequencing costs.
• Improved transcriptome resolution: Enhances accuracy and consistency of sample assignment to improve resolution of individual cell transcriptomes.
• Validation on reference datasets: Demonstrated accuracy and consistency across well-behaved and poorly behaved input data using two well-characterized reference datasets.

Methodology:

The method centers on Bimodal Flexible Fitting (BFF) algorithms (BFFcluster and BFFraw) that classify droplets by modeling bimodal barcode count distributions.